Context: Scalp electroencephalography (EEG) is the gold standard method for recording interictal and ictal activity in epileptic patients. Dynamic and topography of EEG signals are patient-specific. They depends on the individual's brain anatomy, type of the lesion, propagation of the epileptic activity within large-scale epileptic networks. How can we interpret epileptiform events from a neurophysiological perspective? To answer this, we developed a whole brain model based on neuronal physiology that can be used to interpret patient-specific EEG signals.Methods. We developed a whole brain neuro-inspired computational model based on neural masses. Each laminar neural mass model (NMM) of the neocortex consists of Glutamatergic and GABAergic (VIP, PV, SST and NGFC subtypes) neurons. Cellular connectivity and kinetics properties of post synaptic potentials were fitted on physiological knowledge. The brain model consists of a large-scale network of 82 interconnected NMM based on the human connectome. Forward problem is used to simulate high density EEG. Results. Comparison with real data allowed us to reproduce IEDs shape with high degree of realism on scalp EEG in focal epilepsy. Simulations produced hypotheses about cell-related and network-related mechanisms underlying the generation of scalp-recorded epileptic spikes. Simulation of absence seizures reveals a major role of the associative subcortical-cortical loop in the generation of the 3 Hz spike-wave patterns. High density EEG recordings in child during absence seizure and sources reconstruction allowed us to reproduce the individual specific patterns. Conclusion. Physiological models have a high face, integrative and explanatory value that can be tuned to be patient specific.